The Miettinen-Lasota scoring system is the cornestone for prognostic stratification of gastrointestinal stromal tumors(GISTs), based on mitotic count and tumor size. Guidelines lack indications about the optimal sample that should be used for mitoticcount assesment, only specifying that at least 50 high-power fields (HPF) must be evaluated. It is generally believed that, due totumor heterogeneity and the limited sample size obtained, endoscopic ultrasound-guided tissue acquisition (EUS-TA) is unreliable,particularly when using traditional cytology needles. However, only few data are available regarding the use of histology (FNB) needles in this context. Nevertheless, the ability to correctly stratify patients preoperatively is a critical factor in GIST management. The primary aim of this multicenter retrospective study was to evaluate the concordance between EUS-FNB andsurgical specimens in assessing the GIST prognostic score. Secondary endpoins were the adequacy of FNB samples forimmunohistochemistry (IHC) and mutational analysis.

We included all histologically confirmed GISTs from a prospective registry (NCT02855151). Concordance for mitotic count and tumor size measurements was evaluated using Bland-Altman plots, calculating bias (mean difference between FNB and surgical specimens) and limits of agreement (LoA). Agreement in prognostic score classification was calculated as a percentage.

Among a cohort of 68 GISTs, we included 32 cases in which complete histopathological data from both EUS-FNB and surgical specimens were available. The diagnostic concordance between EUS-FNB and surgical pathology was 100%. IHC was successfully performed in all FNB samples (100%). The mitotic count showed a bias of -1.19 mitoses, with LoA ranging from +3.02 to -5.4. For tumor size, the bias was -0.96 mm, with LoA ranging from +21.28 to -23.2 mm. Miettinen-Lasota score concordance between EUS with FNB and surgery showed 72% agreement (23/32): in 5 cases, EUS underestimated the score, while in 4 cases, it overestimated it. Mutational analysis was performed in 62.5% of the cases.

EUS with new FNB needles appears to be reasonably accurate for preoperative GIST evaluation and is certainly a useful tool for diagnosis and mutational analysis. However, its concordance with surgical findings is still suboptimal in terms of prognostic evaluation. On one hand, performing multiple needle passes, together with fanning technique, could improve the representativeness of tumor heterogeneity and increase the amount of material provided to the pathologist. On the other hand, greater attention should be addressed to tumor size measurement in EUS, as this can sometimes lead to score variations. The study's limitations include its retrospective nature and the heterogeneity of mitotic count assessment. Prospective studies with dedicated protocols are warranted, along with an evaluation of the adequacy of the collected material for mitotic count on 50 HPF, as suggested by the guidelines.